AI Fundamentals· Lesson 3 of 10
- 1What Even Is AI? AI vs. ML vs. Deep Learning6 min
- 2How Models Learn: Training vs. Inference6 min
- 3What Is an LLM?8 min
- 4Tokens — the Currency of Everything7 min
- 5Why AI Output Is Probabilistic (and Why It Changes How You Build)7 min
- 6Hallucination — Why Models Make Things Up7 min
- 7What AI Is Genuinely Good and Bad At — a Capability Map7 min
- 8AI Systems vs. Traditional Software — the Mindset Shift7 min
- 9Where AI Fits in a Product (and Where It Shouldn't)7 min
- 10Responsible AI Basics for Builders7 min
What Is an LLM?
In one line · An LLM is a model trained on huge text to predict the next word — and from that prediction, coherent language emerges.
The LLM (Large Language Model) is the heart of every AI product you'll build. Its core job is simple — predict the next token — yet at scale it's astonishing.
Everything ahead — prompting, RAG, agents — is a way to steer the LLM. Understand it and the rest gets easy.
An LLM is like a colleague who read half the internet and completes any sentence you start with the most plausible continuation. It completes patterns, it doesn't think like you.
Like your phone keyboard's next-word suggestion — but millions of times smarter, tracking whole paragraphs of context.
Type 'The capital of Egypt is' and the LLM predicts 'Cairo' — the most likely continuation from its training.
It splits text into tokens, scores each possible next token, picks one, and repeats. The context window is the most tokens it can see at once.
Architecturally LLMs are built on the Transformer. You don't need the math — just: text in, text out, controlled via the prompt and parameters like temperature.
🧪 Try it: give any LLM a half sentence and watch it complete it. Change the last word and see the prediction shift.
Common mistake: treating the LLM as a reliable fact source. It's a source of plausible language; truth isn't guaranteed — hence RAG later.
Use an LLM for any language task: summarizing, classifying, answering, generating, reformatting.
Don't rely on it alone for exact calculations or sensitive facts without a trusted source alongside.